The broad objective of this research project is to improve the transition of elders through the health and long-term-care systems. The project concerns the frequency with which persons age 65 or older move among postacute (PA) and long-term-care (LTC) settings and the effect of transitions on elders. The aims of the first phase of the project are to: 1. Develop a framework for analyzing transitions among PA and LTC settings, 2. Construct an analytic file that integrates Medicare claims and 1994 National Long-Term Care Survey data on short-stay hospital use and PA and LTC in the two-year period prior to 1994 interviews, 3. Determine the frequency of transition patterns, nationally and for key subgroups, and 4. Determine the frequency of events that could indicate transition problems, where in the care process they occur and who is most affected. The second phase of the project, which would be funded separately, would determine the cause of transition problems and where interventions to reduce problems should be targeted to have the greatest impact. The frequency of different types of transitions and events that could indicate transition problems will be estimated using the longitudinal file to be constructed from Medicare claims and 1994 NLTCS respondent data. Transitions among five PA and LTC settings will be studied: nursing homes, chronic disease hospitals, inpatient rehabilitation facilities, formal home care (i.e., at least one caregiver is paid) and informal home care (i.e., no caregiver is paid). The operational measures of potential transition problems include admission to a short- stay hospital within 30 days of a transition (hospitalizations will be grouped into "avoidable" and all other), return to an institutional setting within 30 days of a transition from an institution to the community, an emergency room visit within 30 days of a transition, and an increase in ADL dependence within 3 months of a transition. The statistical methods to be employed in the first phase of the project, which will be primarily descriptive, include likelihood ratio chi-square tests of the independence of two factors and estimation of log-linear models to test the independence of three or more categorical variables.